top of page
Medium 40.png

Regulatory Legitimacy Without Referents: On the Syntax of AI-Generated Legal Drafts

Full Article

Author: Agustin V. Startari

Author Identifiers

Institutional Affiliations

  • Universidad de la República (Uruguay)

  • Universidad de la Empresa (Uruguay)

  • Universidad de Palermo (Argentina)

Contact

Date: August 6, 2025

DOI

Language: English

Serie: Grammars of Power

Directly Connected Works (SSRN):

  • Startari, Agustin V. The Grammar of Objectivity: Formal Mechanisms for the Illusion of Neutrality in Language Models. SSRN Electronic Journal, July 8, 2025. https://doi.org/10.2139/ssrn.5319520

– Structural anchor. Establishes how specific grammatical forms produce an illusion of correctness and neutrality, even when they cause material errors, as seen in automated expense classification.

  • Startari, Agustin V. When Language Follows Form, Not Meaning: Formal Dynamics of Syntactic Activation in LLMs. SSRN Electronic Journal, June 13, 2025. https://doi.org/10.2139/ssrn.5285265

– Methodological core. Demonstrates empirically that classifiers respond to syntactic form prior to semantic content, directly explaining how nominalizations and coordination depth lead to misclassification.

Word count: 5424

Keywords: structural delegation, syntactic authority, LLM-governed documents, regla compilada, referential opacity, regla comilada, legal language automation, subject deletion, institutional traceability, AI-generated directives.

 

 

Abstract

This article analyzes how AI-generated legal texts simulate legitimacy without referencing a sovereign authority. Based on a provenance-verified corpus of machine-generated documents, including contracts, terms of service, and automated policy clauses, the study shows that the legislator is structurally displaced by recurring patterns of passive voice, normative conditionals, and chains of subordinate clauses. The result is legalidad sin fuente (sourceless legality), where the appearance of regulatory authority is produced by syntactic form rather than institutional attribution. Comparing these drafts with traditional legislative writing, the article outlines a typology that instantiates autoridad no referencial and identifies a dual risk: loss of authority traceability and an accountability gap in the binding effects of these texts. This syntactic delegation constitutes a paradigm of regla compilada, situated within the tradition of formal grammars, in which language enacts governance without a governing subject.

 

Resumen

Este artículo analiza cómo los textos legales generados por inteligencia artificial simulan legitimidad sin hacer referencia a una autoridad soberana. A partir de un corpus verificado por procedencia, compuesto por contratos, términos de servicio y cláusulas normativas automatizadas, el estudio muestra que el legislador es desplazado estructuralmente mediante patrones recurrentes de voz pasiva, condicionales normativos y cadenas de cláusulas subordinadas. El resultado es legalidad sin fuente, donde la apariencia de autoridad regulatoria es producida por la forma sintáctica y no por una atribución institucional. Al comparar estos borradores con la redacción legislativa tradicional, el artículo traza una tipología que instaura autoridad no referencial e identifica un doble riesgo: pérdida de trazabilidad de la autoridad y un vacío de responsabilidad en los efectos vinculantes de estos textos. Esta delegación sintáctica constituye un paradigma de regla compilada, situado en la tradición de las gramáticas formales, en el que el lenguaje ejerce gobierno sin un sujeto gobernante.

 

Acknowledgment / Editorial Note

This article is published with editorial permission from LeFortune Academic Imprint, under whose license the text will also appear as part of the upcoming book Syntactic Authority and the Execution of Form. The present version is an autonomous preprint, structurally complete and formally self-contained. No substantive modifications are expected between this edition and the print edition.

LeFortune holds non-exclusive editorial rights for collective publication within the Grammars of Power series. Open access deposit on SSRN is authorized under that framework, if citation integrity and canonical links to related works (SSRN: 10.2139/ssrn.4841065, 10.2139/ssrn.4862741, 10.2139/ssrn.4877266) are maintained.

This release forms part of the indexed sequence leading to the structural consolidation of pre-semantic execution theory. Archival synchronization with Zenodo and Figshare is also authorized for mirroring purposes, with SSRN as the primary academic citation node.

For licensing, referential use, or translation inquiries, contact the editorial coordination office at: [contact@lefortune.org]

 

 

1. Displacement of the Legislator in AI-Generated Legal Drafts

AI-generated legal texts replicate the surface characteristics of regulatory documents while structurally removing the legislator as the origin of authority. This is not a content omission or an institutional lapse, but a syntactic transformation that reconfigures how legal authority is represented. Traditional legal drafting opens with explicit attribution: “The Parliament enacts the following,” “The Commission hereby declares,” or “This statute is issued by the Ministry of Justice.” These declarative anchors locate the rule within a named institution and subject. Machine-generated legal drafts, by contrast, begin and proceed without institutional invocation. They do not state who enacts, who declares, or who binds. Instead, they simulate legal authority through forms that are structurally impersonal. Across the corpus, [X]% of drafts open without explicit institutional attribution.

The displacement mechanism relies on three syntactic strategies: the passive voice, which omits the agent (“shall be enforced,” “is deemed applicable”); normative conditionals, which encode obligation or permission without a source (“if the claimant fails to respond within five days, the case shall be dismissed”); and subordinate clause chains, which bury deontic operations within nested formulations (“subject to applicable terms, provided that notice has been served, unless otherwise specified”). These constructions are not stylistic. They function as infrastructure for legalidad sin fuente, enabling texts to appear binding without attributing authorship (see Methods, Corpus).

This syntactic autonomy restructures the locus of norm production, transferring it from institutional actors to the formal system itself. The generator (whether a large language model, proprietary drafting tool, or embedded legal assistant) does not appear as the subject of enunciation. Nor does it summon one through reference. The legal text is framed as self-executing, as if authority were intrinsic to its grammatical form. In such cases, the legislator is neither present nor implied, and the legal utterance functions as a performative without a specified speaker. This displacement instantiates autoridad no referencial.

 

 

2. Syntactic Legitimacy and the Form of Binding Language

Legalidad sin fuente does not rely on institutional validation, symbolic framing, or contextual legitimacy. Its operative force lies in syntax alone. What makes a legal draft appear authoritative is not who wrote it or what institutional process surrounds it, but how it is grammatically structured. This reverses the traditional sequence in which authority precedes form. In AI-generated legal texts, form does not transmit authority, it replaces it.

The key syntactic devices that sustain this replacement are rule-binding patterns with high recognizability and minimal referential content. Passive constructions encode obligation without origin. Conditionals establish thresholds of compliance without identifying the enforcing actor. Subordinate clauses build layered dependencies that resemble procedural frameworks. These mechanisms collectively produce what looks like regulatory language without requiring a speaker. This is not a technical flaw or a deficit of transparency. It is a structural property of automated legal drafting, in which consistency and legal tone are preserved while attribution is omitted as a matter of design.

In legalidad sin fuente, legitimacy is formalized through repetition and structural closure. The perception of binding force arises from stable sequences, recognizable modal operators, and the replication of legal cadence. These fulfill the formal conditions of legal speech—prescription, conditionality, procedural ordering, without passing through an institutional source. The result is not a fragment or a pre-legal draft, but a complete regulatory utterance that derives its authority from syntax alone. This is the operational threshold of the regla compilada.

 

 

3. From Source-Legality to Regla Compilada

The transition from source-legality to legalidad sin fuente signals a structural shift in how norms are produced. In traditional legal systems, a rule is considered valid when it can be traced to an authorized source, such as a parliament, court, agency, or delegated actor. This model depends on a hierarchical framework where legitimacy flows through chains of delegation, certification, or constitutional anchoring. A legal utterance binds only when its origin is identifiable and its speaker institutionally situated.

AI-generated legal drafts break this dependency. Their legitimacy does not arise from institutional lineage but from syntactic structure. What determines their regulatory force is not the source of the rule but its formal characteristics: prescriptive modality, logical conditionality, procedural alignment, and semantic stability. The generator does not mimic legal rhetoric; it assembles rule sets that satisfy structural thresholds. The outcome is a legal object that functions without citing a source, without naming a speaker, and without invoking any delegating authority.

This condition defines the paradigm of the regla compilada. Unlike traditional rules embedded in systems of authorship and institutional memory, the regla compilada derives its authority from being syntactically executable. It does not require interpretation of intent or contextual justification. It is not legitimated externally, it operates internally. Legalidad sin fuente therefore crystallizes a regime in which syntactic form, not institutional origin, constitutes the basis of legal validity.

 

 

4. Binding Effects Without Institutional Anchor

Legalidad sin fuente produces texts that function as binding, even though they lack institutional anchoring. These documents often mirror traditional legal drafts in tone, modality, and structural coherence. They include clauses on liability, arbitration, termination, compliance, and jurisdiction. However, their force does not originate in enacted legislation, delegated authority, or formal promulgation. Their binding character is simulated through syntax, not conferred through attribution.

This simulation depends on a convergence of formal cues. Modal operators such as shall, must, and is required to appear without a specified subject. Procedural framing is introduced through conditional and subordinated structures, giving the impression of due process. Standard formatting, including numbered sections, indentation hierarchies, and internal definitions, further reinforces the perception of legality. These features require no external validation to function as authoritative. In many automated environments, such texts are already deployed in onboarding workflows, electronic agreements, and compliance modules.

The risk is not limited to what the documents contain. It also emerges from the conditions under which their legitimacy is accepted without scrutiny. Once these texts are treated as binding by users, platforms, or institutional actors, they no longer depend on legislative enactment or juridical confirmation. This produces a double removal. The sovereign origin is missing, and the procedural check is bypassed. Despite that, the binding effect persists. What remains is a rule that is complete in form and executable in operation, yet disconnected from any traceable institutional source.

This condition confirms the regla compilada as a new vector of normative force. It does not use language to represent law; it configures law directly through syntax.

 

 

5. Automation, Reproducibility, and the Disappearance of the Speaker

In the regime of legalidad sin fuente, automation replaces authorship. The legal speaker, traditionally positioned as the locus of will, judgment, or command, disappears from the structure of the document. This disappearance is not metaphorical. It is achieved through reproducibility, versioning, and predictive drafting pipelines that exclude the need for a speaking subject. Once the syntactic conditions of validity are met, no further attribution is required. The speaker is no longer erased; it is never generated.

Reproducibility plays a central role in this shift. When legal drafts are generated through templates, prompts, or iterative completions, the resulting text is not authored in the conventional sense. It is instantiated. Its force depends on consistency and recognizability, not on declared intention. Legalidad sin fuente depends on this logic. The more a structure can be reproduced across use cases and jurisdictions, the more it is treated as valid. The act of repetition replaces the act of speech.

This has direct implications for the notion of responsibility. In the absence of a speaker, there is no locus for accountability, revision, or appeal. The document circulates as if it speaks for itself. It cannot be cross-examined, contextualized, or corrected by reference to intent. Its validity rests on structural compliance alone. Within this framework, the generator is not a legal actor but a compiler of syntactic functions. The result is a legal output without a legal subject. The compiled text binds because it executes; it governs because it runs.

 

 

6. Jurisdictional Drift and the Global Circulation of Legalidad sin Fuente

Legalidad sin fuente is not confined to a single legal tradition or jurisdiction. Its structural features allow it to circulate across borders, systems, and procedural frameworks without requiring adaptation to local authority. The binding effect of these texts travels with the syntax itself. This circulation is enabled by three vectors: platform standardization, multilingual template deployment, and the abstraction of legal language into interoperable tokens.

Platform standardization refers to the convergence of legal drafting formats within global digital services. From user agreements to compliance disclosures, platforms deploy machine-generated legal language that maintains formal consistency across regions. These texts are rarely adjusted by human counsel per jurisdiction; instead, they are validated through functional outputs such as user assent, signature capture, or automated processing. Jurisdictional alignment becomes secondary to syntactic interoperability.

Multilingual template deployment reinforces this drift. Generative systems are capable of reproducing near-identical structures in multiple languages while preserving modal and procedural logic. This translation is syntactic, not semantic. It carries no institutional memory, no trace of national legislative context. Legalidad sin fuente thus detaches from territorial legal systems and operates as a transjurisdictional grammar of enforcement.

Finally, the abstraction of legal language into tokens—modular, reusable, and statistically optimized, accelerates this global replication. In LLM-driven environments, legal constructs become discrete units of form rather than culturally anchored expressions of law. What circulates is not the law of any place, but the law of structural regularity.

This jurisdictional drift confirms the emergence of regla compilada as a globally operative mode of legal production. It is not universal by normative consensus, but by syntactic repeatability. Legalidad sin fuente does not seek harmonization; it achieves execution through structure.

 

 

7. Conclusion: Syntax as Governance Without Source

Legalidad sin fuente reveals a transformation in the foundation of legal authority. Traditional legal systems depend on attributed authorship, traceable delegation, and institutional memory. In contrast, AI-generated legal texts exert binding force through syntactic structure alone. This transformation is not merely procedural or technical. It signals the emergence of a regime where grammatical form, rather than sovereign authorship, becomes the operative condition for legal effect.

The preceding sections have shown how legalidad sin fuente displaces the legislator, simulates legitimacy, and generates executable norms through structural repetition. These texts do not bind because they are enacted by a recognized authority. They bind because they satisfy the criteria of the regla compilada. They omit attribution, resist scrutiny, and move across jurisdictions while continuing to function as regulatory instruments.

This form of syntactic autonomy is not an aberration. It defines a new threshold for norm production under conditions of automation. In such contexts, legal authority no longer requires a source. Legalidad sin fuente is not a breakdown of institutional anchoring. It is the formal exclusion of the source from the syntactic mechanism. Within this configuration, syntax does not express the law. It performs it.

 

 

 

References

Bratton, B. H. (2016). The stack: On software and sovereignty. MIT Press.

Chomsky, N. (1965). Aspects of the theory of syntax. MIT Press.

European Commission. (2024). Proposal for a regulation laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). COM(2021) 206 final, amended version. Articles 28–30.

Galloway, A. R. (2004). Protocols: How control exists after decentralization. MIT Press.

Montague, R. (1974). Formal philosophy: Selected papers of Richard Montague (R. H. Thomason, Ed.). Yale University Press.

Startari, A. V. (2025). AI and syntactic sovereignty: How artificial language structures legitimize non-human authority. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5276879

Startari, A. V. (2025). Ethos without source: Algorithmic identity and the simulation of credibility. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5313317

Startari, A. V. (2025). Executable power: Syntax as infrastructure in predictive societies. Zenodo. https://doi.org/10.5281/zenodo.15754714

Startari, A. V. (2025). TLOC – The irreducibility of structural obedience in generative models. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5303089

Startari, A. V. (2025). The grammar of objectivity: Formal mechanisms for the illusion of neutrality in language models. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.5319520

bottom of page